sched: Add missing state chars to TASK_STATE_TO_CHAR_STR
[linux-2.6/kvm.git] / include / linux / sched.h
blob94858df38a87f74cf7e584e2d47bd0525a954ae6
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
4 /*
5 * cloning flags:
6 */
7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 #define CLONE_IO 0x80000000 /* Clone io context */
33 * Scheduling policies
35 #define SCHED_NORMAL 0
36 #define SCHED_FIFO 1
37 #define SCHED_RR 2
38 #define SCHED_BATCH 3
39 /* SCHED_ISO: reserved but not implemented yet */
40 #define SCHED_IDLE 5
41 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
42 #define SCHED_RESET_ON_FORK 0x40000000
44 #ifdef __KERNEL__
46 struct sched_param {
47 int sched_priority;
50 #include <asm/param.h> /* for HZ */
52 #include <linux/capability.h>
53 #include <linux/threads.h>
54 #include <linux/kernel.h>
55 #include <linux/types.h>
56 #include <linux/timex.h>
57 #include <linux/jiffies.h>
58 #include <linux/rbtree.h>
59 #include <linux/thread_info.h>
60 #include <linux/cpumask.h>
61 #include <linux/errno.h>
62 #include <linux/nodemask.h>
63 #include <linux/mm_types.h>
65 #include <asm/system.h>
66 #include <asm/page.h>
67 #include <asm/ptrace.h>
68 #include <asm/cputime.h>
70 #include <linux/smp.h>
71 #include <linux/sem.h>
72 #include <linux/signal.h>
73 #include <linux/path.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rculist.h>
83 #include <linux/rtmutex.h>
85 #include <linux/time.h>
86 #include <linux/param.h>
87 #include <linux/resource.h>
88 #include <linux/timer.h>
89 #include <linux/hrtimer.h>
90 #include <linux/task_io_accounting.h>
91 #include <linux/kobject.h>
92 #include <linux/latencytop.h>
93 #include <linux/cred.h>
95 #include <asm/processor.h>
97 struct exec_domain;
98 struct futex_pi_state;
99 struct robust_list_head;
100 struct bio;
101 struct fs_struct;
102 struct bts_context;
103 struct perf_event_context;
106 * List of flags we want to share for kernel threads,
107 * if only because they are not used by them anyway.
109 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
112 * These are the constant used to fake the fixed-point load-average
113 * counting. Some notes:
114 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
115 * a load-average precision of 10 bits integer + 11 bits fractional
116 * - if you want to count load-averages more often, you need more
117 * precision, or rounding will get you. With 2-second counting freq,
118 * the EXP_n values would be 1981, 2034 and 2043 if still using only
119 * 11 bit fractions.
121 extern unsigned long avenrun[]; /* Load averages */
122 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
124 #define FSHIFT 11 /* nr of bits of precision */
125 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
126 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
127 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
128 #define EXP_5 2014 /* 1/exp(5sec/5min) */
129 #define EXP_15 2037 /* 1/exp(5sec/15min) */
131 #define CALC_LOAD(load,exp,n) \
132 load *= exp; \
133 load += n*(FIXED_1-exp); \
134 load >>= FSHIFT;
136 extern unsigned long total_forks;
137 extern int nr_threads;
138 DECLARE_PER_CPU(unsigned long, process_counts);
139 extern int nr_processes(void);
140 extern unsigned long nr_running(void);
141 extern unsigned long nr_uninterruptible(void);
142 extern unsigned long nr_iowait(void);
143 extern unsigned long nr_iowait_cpu(void);
144 extern unsigned long this_cpu_load(void);
147 extern void calc_global_load(void);
149 extern unsigned long get_parent_ip(unsigned long addr);
151 struct seq_file;
152 struct cfs_rq;
153 struct task_group;
154 #ifdef CONFIG_SCHED_DEBUG
155 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
156 extern void proc_sched_set_task(struct task_struct *p);
157 extern void
158 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
159 #else
160 static inline void
161 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
164 static inline void proc_sched_set_task(struct task_struct *p)
167 static inline void
168 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
171 #endif
174 * Task state bitmask. NOTE! These bits are also
175 * encoded in fs/proc/array.c: get_task_state().
177 * We have two separate sets of flags: task->state
178 * is about runnability, while task->exit_state are
179 * about the task exiting. Confusing, but this way
180 * modifying one set can't modify the other one by
181 * mistake.
183 #define TASK_RUNNING 0
184 #define TASK_INTERRUPTIBLE 1
185 #define TASK_UNINTERRUPTIBLE 2
186 #define __TASK_STOPPED 4
187 #define __TASK_TRACED 8
188 /* in tsk->exit_state */
189 #define EXIT_ZOMBIE 16
190 #define EXIT_DEAD 32
191 /* in tsk->state again */
192 #define TASK_DEAD 64
193 #define TASK_WAKEKILL 128
194 #define TASK_WAKING 256
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
198 /* Convenience macros for the sake of set_task_state */
199 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
200 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
201 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
203 /* Convenience macros for the sake of wake_up */
204 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
205 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
207 /* get_task_state() */
208 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
209 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
210 __TASK_TRACED)
212 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
213 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
214 #define task_is_stopped_or_traced(task) \
215 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
216 #define task_contributes_to_load(task) \
217 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
218 (task->flags & PF_FREEZING) == 0)
220 #define __set_task_state(tsk, state_value) \
221 do { (tsk)->state = (state_value); } while (0)
222 #define set_task_state(tsk, state_value) \
223 set_mb((tsk)->state, (state_value))
226 * set_current_state() includes a barrier so that the write of current->state
227 * is correctly serialised wrt the caller's subsequent test of whether to
228 * actually sleep:
230 * set_current_state(TASK_UNINTERRUPTIBLE);
231 * if (do_i_need_to_sleep())
232 * schedule();
234 * If the caller does not need such serialisation then use __set_current_state()
236 #define __set_current_state(state_value) \
237 do { current->state = (state_value); } while (0)
238 #define set_current_state(state_value) \
239 set_mb(current->state, (state_value))
241 /* Task command name length */
242 #define TASK_COMM_LEN 16
244 #include <linux/spinlock.h>
247 * This serializes "schedule()" and also protects
248 * the run-queue from deletions/modifications (but
249 * _adding_ to the beginning of the run-queue has
250 * a separate lock).
252 extern rwlock_t tasklist_lock;
253 extern spinlock_t mmlist_lock;
255 struct task_struct;
257 extern void sched_init(void);
258 extern void sched_init_smp(void);
259 extern asmlinkage void schedule_tail(struct task_struct *prev);
260 extern void init_idle(struct task_struct *idle, int cpu);
261 extern void init_idle_bootup_task(struct task_struct *idle);
263 extern int runqueue_is_locked(int cpu);
264 extern void task_rq_unlock_wait(struct task_struct *p);
266 extern cpumask_var_t nohz_cpu_mask;
267 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
268 extern int select_nohz_load_balancer(int cpu);
269 extern int get_nohz_load_balancer(void);
270 #else
271 static inline int select_nohz_load_balancer(int cpu)
273 return 0;
275 #endif
278 * Only dump TASK_* tasks. (0 for all tasks)
280 extern void show_state_filter(unsigned long state_filter);
282 static inline void show_state(void)
284 show_state_filter(0);
287 extern void show_regs(struct pt_regs *);
290 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
291 * task), SP is the stack pointer of the first frame that should be shown in the back
292 * trace (or NULL if the entire call-chain of the task should be shown).
294 extern void show_stack(struct task_struct *task, unsigned long *sp);
296 void io_schedule(void);
297 long io_schedule_timeout(long timeout);
299 extern void cpu_init (void);
300 extern void trap_init(void);
301 extern void update_process_times(int user);
302 extern void scheduler_tick(void);
304 extern void sched_show_task(struct task_struct *p);
306 #ifdef CONFIG_DETECT_SOFTLOCKUP
307 extern void softlockup_tick(void);
308 extern void touch_softlockup_watchdog(void);
309 extern void touch_all_softlockup_watchdogs(void);
310 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
311 void __user *buffer,
312 size_t *lenp, loff_t *ppos);
313 extern unsigned int softlockup_panic;
314 extern int softlockup_thresh;
315 #else
316 static inline void softlockup_tick(void)
319 static inline void touch_softlockup_watchdog(void)
322 static inline void touch_all_softlockup_watchdogs(void)
325 #endif
327 #ifdef CONFIG_DETECT_HUNG_TASK
328 extern unsigned int sysctl_hung_task_panic;
329 extern unsigned long sysctl_hung_task_check_count;
330 extern unsigned long sysctl_hung_task_timeout_secs;
331 extern unsigned long sysctl_hung_task_warnings;
332 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
333 void __user *buffer,
334 size_t *lenp, loff_t *ppos);
335 #endif
337 /* Attach to any functions which should be ignored in wchan output. */
338 #define __sched __attribute__((__section__(".sched.text")))
340 /* Linker adds these: start and end of __sched functions */
341 extern char __sched_text_start[], __sched_text_end[];
343 /* Is this address in the __sched functions? */
344 extern int in_sched_functions(unsigned long addr);
346 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
347 extern signed long schedule_timeout(signed long timeout);
348 extern signed long schedule_timeout_interruptible(signed long timeout);
349 extern signed long schedule_timeout_killable(signed long timeout);
350 extern signed long schedule_timeout_uninterruptible(signed long timeout);
351 asmlinkage void schedule(void);
352 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
354 struct nsproxy;
355 struct user_namespace;
358 * Default maximum number of active map areas, this limits the number of vmas
359 * per mm struct. Users can overwrite this number by sysctl but there is a
360 * problem.
362 * When a program's coredump is generated as ELF format, a section is created
363 * per a vma. In ELF, the number of sections is represented in unsigned short.
364 * This means the number of sections should be smaller than 65535 at coredump.
365 * Because the kernel adds some informative sections to a image of program at
366 * generating coredump, we need some margin. The number of extra sections is
367 * 1-3 now and depends on arch. We use "5" as safe margin, here.
369 #define MAPCOUNT_ELF_CORE_MARGIN (5)
370 #define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
372 extern int sysctl_max_map_count;
374 #include <linux/aio.h>
376 extern unsigned long
377 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
378 unsigned long, unsigned long);
379 extern unsigned long
380 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
381 unsigned long len, unsigned long pgoff,
382 unsigned long flags);
383 extern void arch_unmap_area(struct mm_struct *, unsigned long);
384 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
386 #if USE_SPLIT_PTLOCKS
388 * The mm counters are not protected by its page_table_lock,
389 * so must be incremented atomically.
391 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
392 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
393 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
394 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
395 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
397 #else /* !USE_SPLIT_PTLOCKS */
399 * The mm counters are protected by its page_table_lock,
400 * so can be incremented directly.
402 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
403 #define get_mm_counter(mm, member) ((mm)->_##member)
404 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
405 #define inc_mm_counter(mm, member) (mm)->_##member++
406 #define dec_mm_counter(mm, member) (mm)->_##member--
408 #endif /* !USE_SPLIT_PTLOCKS */
410 #define get_mm_rss(mm) \
411 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
412 #define update_hiwater_rss(mm) do { \
413 unsigned long _rss = get_mm_rss(mm); \
414 if ((mm)->hiwater_rss < _rss) \
415 (mm)->hiwater_rss = _rss; \
416 } while (0)
417 #define update_hiwater_vm(mm) do { \
418 if ((mm)->hiwater_vm < (mm)->total_vm) \
419 (mm)->hiwater_vm = (mm)->total_vm; \
420 } while (0)
422 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
424 return max(mm->hiwater_rss, get_mm_rss(mm));
427 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
428 struct mm_struct *mm)
430 unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
432 if (*maxrss < hiwater_rss)
433 *maxrss = hiwater_rss;
436 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
438 return max(mm->hiwater_vm, mm->total_vm);
441 extern void set_dumpable(struct mm_struct *mm, int value);
442 extern int get_dumpable(struct mm_struct *mm);
444 /* mm flags */
445 /* dumpable bits */
446 #define MMF_DUMPABLE 0 /* core dump is permitted */
447 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
449 #define MMF_DUMPABLE_BITS 2
450 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
452 /* coredump filter bits */
453 #define MMF_DUMP_ANON_PRIVATE 2
454 #define MMF_DUMP_ANON_SHARED 3
455 #define MMF_DUMP_MAPPED_PRIVATE 4
456 #define MMF_DUMP_MAPPED_SHARED 5
457 #define MMF_DUMP_ELF_HEADERS 6
458 #define MMF_DUMP_HUGETLB_PRIVATE 7
459 #define MMF_DUMP_HUGETLB_SHARED 8
461 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
462 #define MMF_DUMP_FILTER_BITS 7
463 #define MMF_DUMP_FILTER_MASK \
464 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
465 #define MMF_DUMP_FILTER_DEFAULT \
466 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
467 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
469 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
470 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
471 #else
472 # define MMF_DUMP_MASK_DEFAULT_ELF 0
473 #endif
474 /* leave room for more dump flags */
475 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
477 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
479 struct sighand_struct {
480 atomic_t count;
481 struct k_sigaction action[_NSIG];
482 spinlock_t siglock;
483 wait_queue_head_t signalfd_wqh;
486 struct pacct_struct {
487 int ac_flag;
488 long ac_exitcode;
489 unsigned long ac_mem;
490 cputime_t ac_utime, ac_stime;
491 unsigned long ac_minflt, ac_majflt;
494 struct cpu_itimer {
495 cputime_t expires;
496 cputime_t incr;
497 u32 error;
498 u32 incr_error;
502 * struct task_cputime - collected CPU time counts
503 * @utime: time spent in user mode, in &cputime_t units
504 * @stime: time spent in kernel mode, in &cputime_t units
505 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
507 * This structure groups together three kinds of CPU time that are
508 * tracked for threads and thread groups. Most things considering
509 * CPU time want to group these counts together and treat all three
510 * of them in parallel.
512 struct task_cputime {
513 cputime_t utime;
514 cputime_t stime;
515 unsigned long long sum_exec_runtime;
517 /* Alternate field names when used to cache expirations. */
518 #define prof_exp stime
519 #define virt_exp utime
520 #define sched_exp sum_exec_runtime
522 #define INIT_CPUTIME \
523 (struct task_cputime) { \
524 .utime = cputime_zero, \
525 .stime = cputime_zero, \
526 .sum_exec_runtime = 0, \
530 * Disable preemption until the scheduler is running.
531 * Reset by start_kernel()->sched_init()->init_idle().
533 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
534 * before the scheduler is active -- see should_resched().
536 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
539 * struct thread_group_cputimer - thread group interval timer counts
540 * @cputime: thread group interval timers.
541 * @running: non-zero when there are timers running and
542 * @cputime receives updates.
543 * @lock: lock for fields in this struct.
545 * This structure contains the version of task_cputime, above, that is
546 * used for thread group CPU timer calculations.
548 struct thread_group_cputimer {
549 struct task_cputime cputime;
550 int running;
551 spinlock_t lock;
555 * NOTE! "signal_struct" does not have it's own
556 * locking, because a shared signal_struct always
557 * implies a shared sighand_struct, so locking
558 * sighand_struct is always a proper superset of
559 * the locking of signal_struct.
561 struct signal_struct {
562 atomic_t count;
563 atomic_t live;
565 wait_queue_head_t wait_chldexit; /* for wait4() */
567 /* current thread group signal load-balancing target: */
568 struct task_struct *curr_target;
570 /* shared signal handling: */
571 struct sigpending shared_pending;
573 /* thread group exit support */
574 int group_exit_code;
575 /* overloaded:
576 * - notify group_exit_task when ->count is equal to notify_count
577 * - everyone except group_exit_task is stopped during signal delivery
578 * of fatal signals, group_exit_task processes the signal.
580 int notify_count;
581 struct task_struct *group_exit_task;
583 /* thread group stop support, overloads group_exit_code too */
584 int group_stop_count;
585 unsigned int flags; /* see SIGNAL_* flags below */
587 /* POSIX.1b Interval Timers */
588 struct list_head posix_timers;
590 /* ITIMER_REAL timer for the process */
591 struct hrtimer real_timer;
592 struct pid *leader_pid;
593 ktime_t it_real_incr;
596 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
597 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
598 * values are defined to 0 and 1 respectively
600 struct cpu_itimer it[2];
603 * Thread group totals for process CPU timers.
604 * See thread_group_cputimer(), et al, for details.
606 struct thread_group_cputimer cputimer;
608 /* Earliest-expiration cache. */
609 struct task_cputime cputime_expires;
611 struct list_head cpu_timers[3];
613 struct pid *tty_old_pgrp;
615 /* boolean value for session group leader */
616 int leader;
618 struct tty_struct *tty; /* NULL if no tty */
621 * Cumulative resource counters for dead threads in the group,
622 * and for reaped dead child processes forked by this group.
623 * Live threads maintain their own counters and add to these
624 * in __exit_signal, except for the group leader.
626 cputime_t utime, stime, cutime, cstime;
627 cputime_t gtime;
628 cputime_t cgtime;
629 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
630 cputime_t prev_utime, prev_stime;
631 #endif
632 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
633 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
634 unsigned long inblock, oublock, cinblock, coublock;
635 unsigned long maxrss, cmaxrss;
636 struct task_io_accounting ioac;
639 * Cumulative ns of schedule CPU time fo dead threads in the
640 * group, not including a zombie group leader, (This only differs
641 * from jiffies_to_ns(utime + stime) if sched_clock uses something
642 * other than jiffies.)
644 unsigned long long sum_sched_runtime;
647 * We don't bother to synchronize most readers of this at all,
648 * because there is no reader checking a limit that actually needs
649 * to get both rlim_cur and rlim_max atomically, and either one
650 * alone is a single word that can safely be read normally.
651 * getrlimit/setrlimit use task_lock(current->group_leader) to
652 * protect this instead of the siglock, because they really
653 * have no need to disable irqs.
655 struct rlimit rlim[RLIM_NLIMITS];
657 #ifdef CONFIG_BSD_PROCESS_ACCT
658 struct pacct_struct pacct; /* per-process accounting information */
659 #endif
660 #ifdef CONFIG_TASKSTATS
661 struct taskstats *stats;
662 #endif
663 #ifdef CONFIG_AUDIT
664 unsigned audit_tty;
665 struct tty_audit_buf *tty_audit_buf;
666 #endif
668 int oom_adj; /* OOM kill score adjustment (bit shift) */
671 /* Context switch must be unlocked if interrupts are to be enabled */
672 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
673 # define __ARCH_WANT_UNLOCKED_CTXSW
674 #endif
677 * Bits in flags field of signal_struct.
679 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
680 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
681 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
682 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
684 * Pending notifications to parent.
686 #define SIGNAL_CLD_STOPPED 0x00000010
687 #define SIGNAL_CLD_CONTINUED 0x00000020
688 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
690 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
692 /* If true, all threads except ->group_exit_task have pending SIGKILL */
693 static inline int signal_group_exit(const struct signal_struct *sig)
695 return (sig->flags & SIGNAL_GROUP_EXIT) ||
696 (sig->group_exit_task != NULL);
700 * Some day this will be a full-fledged user tracking system..
702 struct user_struct {
703 atomic_t __count; /* reference count */
704 atomic_t processes; /* How many processes does this user have? */
705 atomic_t files; /* How many open files does this user have? */
706 atomic_t sigpending; /* How many pending signals does this user have? */
707 #ifdef CONFIG_INOTIFY_USER
708 atomic_t inotify_watches; /* How many inotify watches does this user have? */
709 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
710 #endif
711 #ifdef CONFIG_EPOLL
712 atomic_t epoll_watches; /* The number of file descriptors currently watched */
713 #endif
714 #ifdef CONFIG_POSIX_MQUEUE
715 /* protected by mq_lock */
716 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
717 #endif
718 unsigned long locked_shm; /* How many pages of mlocked shm ? */
720 #ifdef CONFIG_KEYS
721 struct key *uid_keyring; /* UID specific keyring */
722 struct key *session_keyring; /* UID's default session keyring */
723 #endif
725 /* Hash table maintenance information */
726 struct hlist_node uidhash_node;
727 uid_t uid;
728 struct user_namespace *user_ns;
730 #ifdef CONFIG_USER_SCHED
731 struct task_group *tg;
732 #ifdef CONFIG_SYSFS
733 struct kobject kobj;
734 struct delayed_work work;
735 #endif
736 #endif
738 #ifdef CONFIG_PERF_EVENTS
739 atomic_long_t locked_vm;
740 #endif
743 extern int uids_sysfs_init(void);
745 extern struct user_struct *find_user(uid_t);
747 extern struct user_struct root_user;
748 #define INIT_USER (&root_user)
751 struct backing_dev_info;
752 struct reclaim_state;
754 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
755 struct sched_info {
756 /* cumulative counters */
757 unsigned long pcount; /* # of times run on this cpu */
758 unsigned long long run_delay; /* time spent waiting on a runqueue */
760 /* timestamps */
761 unsigned long long last_arrival,/* when we last ran on a cpu */
762 last_queued; /* when we were last queued to run */
763 #ifdef CONFIG_SCHEDSTATS
764 /* BKL stats */
765 unsigned int bkl_count;
766 #endif
768 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
770 #ifdef CONFIG_TASK_DELAY_ACCT
771 struct task_delay_info {
772 spinlock_t lock;
773 unsigned int flags; /* Private per-task flags */
775 /* For each stat XXX, add following, aligned appropriately
777 * struct timespec XXX_start, XXX_end;
778 * u64 XXX_delay;
779 * u32 XXX_count;
781 * Atomicity of updates to XXX_delay, XXX_count protected by
782 * single lock above (split into XXX_lock if contention is an issue).
786 * XXX_count is incremented on every XXX operation, the delay
787 * associated with the operation is added to XXX_delay.
788 * XXX_delay contains the accumulated delay time in nanoseconds.
790 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
791 u64 blkio_delay; /* wait for sync block io completion */
792 u64 swapin_delay; /* wait for swapin block io completion */
793 u32 blkio_count; /* total count of the number of sync block */
794 /* io operations performed */
795 u32 swapin_count; /* total count of the number of swapin block */
796 /* io operations performed */
798 struct timespec freepages_start, freepages_end;
799 u64 freepages_delay; /* wait for memory reclaim */
800 u32 freepages_count; /* total count of memory reclaim */
802 #endif /* CONFIG_TASK_DELAY_ACCT */
804 static inline int sched_info_on(void)
806 #ifdef CONFIG_SCHEDSTATS
807 return 1;
808 #elif defined(CONFIG_TASK_DELAY_ACCT)
809 extern int delayacct_on;
810 return delayacct_on;
811 #else
812 return 0;
813 #endif
816 enum cpu_idle_type {
817 CPU_IDLE,
818 CPU_NOT_IDLE,
819 CPU_NEWLY_IDLE,
820 CPU_MAX_IDLE_TYPES
824 * sched-domains (multiprocessor balancing) declarations:
828 * Increase resolution of nice-level calculations:
830 #define SCHED_LOAD_SHIFT 10
831 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
833 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
835 #ifdef CONFIG_SMP
836 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
837 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
838 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
839 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
840 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
841 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
842 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
843 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
844 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
845 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
846 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
848 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
850 enum powersavings_balance_level {
851 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
852 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
853 * first for long running threads
855 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
856 * cpu package for power savings
858 MAX_POWERSAVINGS_BALANCE_LEVELS
861 extern int sched_mc_power_savings, sched_smt_power_savings;
863 static inline int sd_balance_for_mc_power(void)
865 if (sched_smt_power_savings)
866 return SD_POWERSAVINGS_BALANCE;
868 return SD_PREFER_SIBLING;
871 static inline int sd_balance_for_package_power(void)
873 if (sched_mc_power_savings | sched_smt_power_savings)
874 return SD_POWERSAVINGS_BALANCE;
876 return SD_PREFER_SIBLING;
880 * Optimise SD flags for power savings:
881 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
882 * Keep default SD flags if sched_{smt,mc}_power_saving=0
885 static inline int sd_power_saving_flags(void)
887 if (sched_mc_power_savings | sched_smt_power_savings)
888 return SD_BALANCE_NEWIDLE;
890 return 0;
893 struct sched_group {
894 struct sched_group *next; /* Must be a circular list */
897 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
898 * single CPU.
900 unsigned int cpu_power;
903 * The CPUs this group covers.
905 * NOTE: this field is variable length. (Allocated dynamically
906 * by attaching extra space to the end of the structure,
907 * depending on how many CPUs the kernel has booted up with)
909 * It is also be embedded into static data structures at build
910 * time. (See 'struct static_sched_group' in kernel/sched.c)
912 unsigned long cpumask[0];
915 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
917 return to_cpumask(sg->cpumask);
920 enum sched_domain_level {
921 SD_LV_NONE = 0,
922 SD_LV_SIBLING,
923 SD_LV_MC,
924 SD_LV_CPU,
925 SD_LV_NODE,
926 SD_LV_ALLNODES,
927 SD_LV_MAX
930 struct sched_domain_attr {
931 int relax_domain_level;
934 #define SD_ATTR_INIT (struct sched_domain_attr) { \
935 .relax_domain_level = -1, \
938 struct sched_domain {
939 /* These fields must be setup */
940 struct sched_domain *parent; /* top domain must be null terminated */
941 struct sched_domain *child; /* bottom domain must be null terminated */
942 struct sched_group *groups; /* the balancing groups of the domain */
943 unsigned long min_interval; /* Minimum balance interval ms */
944 unsigned long max_interval; /* Maximum balance interval ms */
945 unsigned int busy_factor; /* less balancing by factor if busy */
946 unsigned int imbalance_pct; /* No balance until over watermark */
947 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
948 unsigned int busy_idx;
949 unsigned int idle_idx;
950 unsigned int newidle_idx;
951 unsigned int wake_idx;
952 unsigned int forkexec_idx;
953 unsigned int smt_gain;
954 int flags; /* See SD_* */
955 enum sched_domain_level level;
957 /* Runtime fields. */
958 unsigned long last_balance; /* init to jiffies. units in jiffies */
959 unsigned int balance_interval; /* initialise to 1. units in ms. */
960 unsigned int nr_balance_failed; /* initialise to 0 */
962 u64 last_update;
964 #ifdef CONFIG_SCHEDSTATS
965 /* load_balance() stats */
966 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
967 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
968 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
969 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
970 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
971 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
972 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
973 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
975 /* Active load balancing */
976 unsigned int alb_count;
977 unsigned int alb_failed;
978 unsigned int alb_pushed;
980 /* SD_BALANCE_EXEC stats */
981 unsigned int sbe_count;
982 unsigned int sbe_balanced;
983 unsigned int sbe_pushed;
985 /* SD_BALANCE_FORK stats */
986 unsigned int sbf_count;
987 unsigned int sbf_balanced;
988 unsigned int sbf_pushed;
990 /* try_to_wake_up() stats */
991 unsigned int ttwu_wake_remote;
992 unsigned int ttwu_move_affine;
993 unsigned int ttwu_move_balance;
994 #endif
995 #ifdef CONFIG_SCHED_DEBUG
996 char *name;
997 #endif
1000 * Span of all CPUs in this domain.
1002 * NOTE: this field is variable length. (Allocated dynamically
1003 * by attaching extra space to the end of the structure,
1004 * depending on how many CPUs the kernel has booted up with)
1006 * It is also be embedded into static data structures at build
1007 * time. (See 'struct static_sched_domain' in kernel/sched.c)
1009 unsigned long span[0];
1012 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1014 return to_cpumask(sd->span);
1017 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1018 struct sched_domain_attr *dattr_new);
1020 /* Allocate an array of sched domains, for partition_sched_domains(). */
1021 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1022 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1024 /* Test a flag in parent sched domain */
1025 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1027 if (sd->parent && (sd->parent->flags & flag))
1028 return 1;
1030 return 0;
1033 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1034 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1036 #else /* CONFIG_SMP */
1038 struct sched_domain_attr;
1040 static inline void
1041 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1042 struct sched_domain_attr *dattr_new)
1045 #endif /* !CONFIG_SMP */
1048 struct io_context; /* See blkdev.h */
1051 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1052 extern void prefetch_stack(struct task_struct *t);
1053 #else
1054 static inline void prefetch_stack(struct task_struct *t) { }
1055 #endif
1057 struct audit_context; /* See audit.c */
1058 struct mempolicy;
1059 struct pipe_inode_info;
1060 struct uts_namespace;
1062 struct rq;
1063 struct sched_domain;
1066 * wake flags
1068 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1069 #define WF_FORK 0x02 /* child wakeup after fork */
1071 struct sched_class {
1072 const struct sched_class *next;
1074 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1075 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1076 void (*yield_task) (struct rq *rq);
1078 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1080 struct task_struct * (*pick_next_task) (struct rq *rq);
1081 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1083 #ifdef CONFIG_SMP
1084 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1086 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1087 struct rq *busiest, unsigned long max_load_move,
1088 struct sched_domain *sd, enum cpu_idle_type idle,
1089 int *all_pinned, int *this_best_prio);
1091 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1092 struct rq *busiest, struct sched_domain *sd,
1093 enum cpu_idle_type idle);
1094 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1095 void (*post_schedule) (struct rq *this_rq);
1096 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1097 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1099 void (*set_cpus_allowed)(struct task_struct *p,
1100 const struct cpumask *newmask);
1102 void (*rq_online)(struct rq *rq);
1103 void (*rq_offline)(struct rq *rq);
1104 #endif
1106 void (*set_curr_task) (struct rq *rq);
1107 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1108 void (*task_fork) (struct task_struct *p);
1110 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1111 int running);
1112 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1113 int running);
1114 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1115 int oldprio, int running);
1117 unsigned int (*get_rr_interval) (struct rq *rq,
1118 struct task_struct *task);
1120 #ifdef CONFIG_FAIR_GROUP_SCHED
1121 void (*moved_group) (struct task_struct *p, int on_rq);
1122 #endif
1125 struct load_weight {
1126 unsigned long weight, inv_weight;
1130 * CFS stats for a schedulable entity (task, task-group etc)
1132 * Current field usage histogram:
1134 * 4 se->block_start
1135 * 4 se->run_node
1136 * 4 se->sleep_start
1137 * 6 se->load.weight
1139 struct sched_entity {
1140 struct load_weight load; /* for load-balancing */
1141 struct rb_node run_node;
1142 struct list_head group_node;
1143 unsigned int on_rq;
1145 u64 exec_start;
1146 u64 sum_exec_runtime;
1147 u64 vruntime;
1148 u64 prev_sum_exec_runtime;
1150 u64 last_wakeup;
1151 u64 avg_overlap;
1153 u64 nr_migrations;
1155 u64 start_runtime;
1156 u64 avg_wakeup;
1158 #ifdef CONFIG_SCHEDSTATS
1159 u64 wait_start;
1160 u64 wait_max;
1161 u64 wait_count;
1162 u64 wait_sum;
1163 u64 iowait_count;
1164 u64 iowait_sum;
1166 u64 sleep_start;
1167 u64 sleep_max;
1168 s64 sum_sleep_runtime;
1170 u64 block_start;
1171 u64 block_max;
1172 u64 exec_max;
1173 u64 slice_max;
1175 u64 nr_migrations_cold;
1176 u64 nr_failed_migrations_affine;
1177 u64 nr_failed_migrations_running;
1178 u64 nr_failed_migrations_hot;
1179 u64 nr_forced_migrations;
1181 u64 nr_wakeups;
1182 u64 nr_wakeups_sync;
1183 u64 nr_wakeups_migrate;
1184 u64 nr_wakeups_local;
1185 u64 nr_wakeups_remote;
1186 u64 nr_wakeups_affine;
1187 u64 nr_wakeups_affine_attempts;
1188 u64 nr_wakeups_passive;
1189 u64 nr_wakeups_idle;
1190 #endif
1192 #ifdef CONFIG_FAIR_GROUP_SCHED
1193 struct sched_entity *parent;
1194 /* rq on which this entity is (to be) queued: */
1195 struct cfs_rq *cfs_rq;
1196 /* rq "owned" by this entity/group: */
1197 struct cfs_rq *my_q;
1198 #endif
1201 struct sched_rt_entity {
1202 struct list_head run_list;
1203 unsigned long timeout;
1204 unsigned int time_slice;
1205 int nr_cpus_allowed;
1207 struct sched_rt_entity *back;
1208 #ifdef CONFIG_RT_GROUP_SCHED
1209 struct sched_rt_entity *parent;
1210 /* rq on which this entity is (to be) queued: */
1211 struct rt_rq *rt_rq;
1212 /* rq "owned" by this entity/group: */
1213 struct rt_rq *my_q;
1214 #endif
1217 struct rcu_node;
1219 struct task_struct {
1220 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1221 void *stack;
1222 atomic_t usage;
1223 unsigned int flags; /* per process flags, defined below */
1224 unsigned int ptrace;
1226 int lock_depth; /* BKL lock depth */
1228 #ifdef CONFIG_SMP
1229 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1230 int oncpu;
1231 #endif
1232 #endif
1234 int prio, static_prio, normal_prio;
1235 unsigned int rt_priority;
1236 const struct sched_class *sched_class;
1237 struct sched_entity se;
1238 struct sched_rt_entity rt;
1240 #ifdef CONFIG_PREEMPT_NOTIFIERS
1241 /* list of struct preempt_notifier: */
1242 struct hlist_head preempt_notifiers;
1243 #endif
1246 * fpu_counter contains the number of consecutive context switches
1247 * that the FPU is used. If this is over a threshold, the lazy fpu
1248 * saving becomes unlazy to save the trap. This is an unsigned char
1249 * so that after 256 times the counter wraps and the behavior turns
1250 * lazy again; this to deal with bursty apps that only use FPU for
1251 * a short time
1253 unsigned char fpu_counter;
1254 #ifdef CONFIG_BLK_DEV_IO_TRACE
1255 unsigned int btrace_seq;
1256 #endif
1258 unsigned int policy;
1259 cpumask_t cpus_allowed;
1261 #ifdef CONFIG_TREE_PREEMPT_RCU
1262 int rcu_read_lock_nesting;
1263 char rcu_read_unlock_special;
1264 struct rcu_node *rcu_blocked_node;
1265 struct list_head rcu_node_entry;
1266 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1268 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1269 struct sched_info sched_info;
1270 #endif
1272 struct list_head tasks;
1273 struct plist_node pushable_tasks;
1275 struct mm_struct *mm, *active_mm;
1277 /* task state */
1278 int exit_state;
1279 int exit_code, exit_signal;
1280 int pdeath_signal; /* The signal sent when the parent dies */
1281 /* ??? */
1282 unsigned int personality;
1283 unsigned did_exec:1;
1284 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1285 * execve */
1286 unsigned in_iowait:1;
1289 /* Revert to default priority/policy when forking */
1290 unsigned sched_reset_on_fork:1;
1292 pid_t pid;
1293 pid_t tgid;
1295 #ifdef CONFIG_CC_STACKPROTECTOR
1296 /* Canary value for the -fstack-protector gcc feature */
1297 unsigned long stack_canary;
1298 #endif
1301 * pointers to (original) parent process, youngest child, younger sibling,
1302 * older sibling, respectively. (p->father can be replaced with
1303 * p->real_parent->pid)
1305 struct task_struct *real_parent; /* real parent process */
1306 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1308 * children/sibling forms the list of my natural children
1310 struct list_head children; /* list of my children */
1311 struct list_head sibling; /* linkage in my parent's children list */
1312 struct task_struct *group_leader; /* threadgroup leader */
1315 * ptraced is the list of tasks this task is using ptrace on.
1316 * This includes both natural children and PTRACE_ATTACH targets.
1317 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1319 struct list_head ptraced;
1320 struct list_head ptrace_entry;
1323 * This is the tracer handle for the ptrace BTS extension.
1324 * This field actually belongs to the ptracer task.
1326 struct bts_context *bts;
1328 /* PID/PID hash table linkage. */
1329 struct pid_link pids[PIDTYPE_MAX];
1330 struct list_head thread_group;
1332 struct completion *vfork_done; /* for vfork() */
1333 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1334 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1336 cputime_t utime, stime, utimescaled, stimescaled;
1337 cputime_t gtime;
1338 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1339 cputime_t prev_utime, prev_stime;
1340 #endif
1341 unsigned long nvcsw, nivcsw; /* context switch counts */
1342 struct timespec start_time; /* monotonic time */
1343 struct timespec real_start_time; /* boot based time */
1344 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1345 unsigned long min_flt, maj_flt;
1347 struct task_cputime cputime_expires;
1348 struct list_head cpu_timers[3];
1350 /* process credentials */
1351 const struct cred *real_cred; /* objective and real subjective task
1352 * credentials (COW) */
1353 const struct cred *cred; /* effective (overridable) subjective task
1354 * credentials (COW) */
1355 struct mutex cred_guard_mutex; /* guard against foreign influences on
1356 * credential calculations
1357 * (notably. ptrace) */
1358 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1360 char comm[TASK_COMM_LEN]; /* executable name excluding path
1361 - access with [gs]et_task_comm (which lock
1362 it with task_lock())
1363 - initialized normally by flush_old_exec */
1364 /* file system info */
1365 int link_count, total_link_count;
1366 #ifdef CONFIG_SYSVIPC
1367 /* ipc stuff */
1368 struct sysv_sem sysvsem;
1369 #endif
1370 #ifdef CONFIG_DETECT_HUNG_TASK
1371 /* hung task detection */
1372 unsigned long last_switch_count;
1373 #endif
1374 /* CPU-specific state of this task */
1375 struct thread_struct thread;
1376 /* filesystem information */
1377 struct fs_struct *fs;
1378 /* open file information */
1379 struct files_struct *files;
1380 /* namespaces */
1381 struct nsproxy *nsproxy;
1382 /* signal handlers */
1383 struct signal_struct *signal;
1384 struct sighand_struct *sighand;
1386 sigset_t blocked, real_blocked;
1387 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1388 struct sigpending pending;
1390 unsigned long sas_ss_sp;
1391 size_t sas_ss_size;
1392 int (*notifier)(void *priv);
1393 void *notifier_data;
1394 sigset_t *notifier_mask;
1395 struct audit_context *audit_context;
1396 #ifdef CONFIG_AUDITSYSCALL
1397 uid_t loginuid;
1398 unsigned int sessionid;
1399 #endif
1400 seccomp_t seccomp;
1402 /* Thread group tracking */
1403 u32 parent_exec_id;
1404 u32 self_exec_id;
1405 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1406 * mempolicy */
1407 spinlock_t alloc_lock;
1409 #ifdef CONFIG_GENERIC_HARDIRQS
1410 /* IRQ handler threads */
1411 struct irqaction *irqaction;
1412 #endif
1414 /* Protection of the PI data structures: */
1415 raw_spinlock_t pi_lock;
1417 #ifdef CONFIG_RT_MUTEXES
1418 /* PI waiters blocked on a rt_mutex held by this task */
1419 struct plist_head pi_waiters;
1420 /* Deadlock detection and priority inheritance handling */
1421 struct rt_mutex_waiter *pi_blocked_on;
1422 #endif
1424 #ifdef CONFIG_DEBUG_MUTEXES
1425 /* mutex deadlock detection */
1426 struct mutex_waiter *blocked_on;
1427 #endif
1428 #ifdef CONFIG_TRACE_IRQFLAGS
1429 unsigned int irq_events;
1430 unsigned long hardirq_enable_ip;
1431 unsigned long hardirq_disable_ip;
1432 unsigned int hardirq_enable_event;
1433 unsigned int hardirq_disable_event;
1434 int hardirqs_enabled;
1435 int hardirq_context;
1436 unsigned long softirq_disable_ip;
1437 unsigned long softirq_enable_ip;
1438 unsigned int softirq_disable_event;
1439 unsigned int softirq_enable_event;
1440 int softirqs_enabled;
1441 int softirq_context;
1442 #endif
1443 #ifdef CONFIG_LOCKDEP
1444 # define MAX_LOCK_DEPTH 48UL
1445 u64 curr_chain_key;
1446 int lockdep_depth;
1447 unsigned int lockdep_recursion;
1448 struct held_lock held_locks[MAX_LOCK_DEPTH];
1449 gfp_t lockdep_reclaim_gfp;
1450 #endif
1452 #ifdef CONFIG_FS_JOURNAL_INFO
1453 /* journalling filesystem info */
1454 void *journal_info;
1455 #endif
1457 /* stacked block device info */
1458 struct bio *bio_list, **bio_tail;
1460 /* VM state */
1461 struct reclaim_state *reclaim_state;
1463 struct backing_dev_info *backing_dev_info;
1465 struct io_context *io_context;
1467 unsigned long ptrace_message;
1468 siginfo_t *last_siginfo; /* For ptrace use. */
1469 struct task_io_accounting ioac;
1470 #if defined(CONFIG_TASK_XACCT)
1471 u64 acct_rss_mem1; /* accumulated rss usage */
1472 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1473 cputime_t acct_timexpd; /* stime + utime since last update */
1474 #endif
1475 #ifdef CONFIG_CPUSETS
1476 nodemask_t mems_allowed; /* Protected by alloc_lock */
1477 int cpuset_mem_spread_rotor;
1478 #endif
1479 #ifdef CONFIG_CGROUPS
1480 /* Control Group info protected by css_set_lock */
1481 struct css_set *cgroups;
1482 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1483 struct list_head cg_list;
1484 #endif
1485 #ifdef CONFIG_FUTEX
1486 struct robust_list_head __user *robust_list;
1487 #ifdef CONFIG_COMPAT
1488 struct compat_robust_list_head __user *compat_robust_list;
1489 #endif
1490 struct list_head pi_state_list;
1491 struct futex_pi_state *pi_state_cache;
1492 #endif
1493 #ifdef CONFIG_PERF_EVENTS
1494 struct perf_event_context *perf_event_ctxp;
1495 struct mutex perf_event_mutex;
1496 struct list_head perf_event_list;
1497 #endif
1498 #ifdef CONFIG_NUMA
1499 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1500 short il_next;
1501 #endif
1502 atomic_t fs_excl; /* holding fs exclusive resources */
1503 struct rcu_head rcu;
1506 * cache last used pipe for splice
1508 struct pipe_inode_info *splice_pipe;
1509 #ifdef CONFIG_TASK_DELAY_ACCT
1510 struct task_delay_info *delays;
1511 #endif
1512 #ifdef CONFIG_FAULT_INJECTION
1513 int make_it_fail;
1514 #endif
1515 struct prop_local_single dirties;
1516 #ifdef CONFIG_LATENCYTOP
1517 int latency_record_count;
1518 struct latency_record latency_record[LT_SAVECOUNT];
1519 #endif
1521 * time slack values; these are used to round up poll() and
1522 * select() etc timeout values. These are in nanoseconds.
1524 unsigned long timer_slack_ns;
1525 unsigned long default_timer_slack_ns;
1527 struct list_head *scm_work_list;
1528 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1529 /* Index of current stored adress in ret_stack */
1530 int curr_ret_stack;
1531 /* Stack of return addresses for return function tracing */
1532 struct ftrace_ret_stack *ret_stack;
1533 /* time stamp for last schedule */
1534 unsigned long long ftrace_timestamp;
1536 * Number of functions that haven't been traced
1537 * because of depth overrun.
1539 atomic_t trace_overrun;
1540 /* Pause for the tracing */
1541 atomic_t tracing_graph_pause;
1542 #endif
1543 #ifdef CONFIG_TRACING
1544 /* state flags for use by tracers */
1545 unsigned long trace;
1546 /* bitmask of trace recursion */
1547 unsigned long trace_recursion;
1548 #endif /* CONFIG_TRACING */
1549 unsigned long stack_start;
1552 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1553 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1556 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1557 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1558 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1559 * values are inverted: lower p->prio value means higher priority.
1561 * The MAX_USER_RT_PRIO value allows the actual maximum
1562 * RT priority to be separate from the value exported to
1563 * user-space. This allows kernel threads to set their
1564 * priority to a value higher than any user task. Note:
1565 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1568 #define MAX_USER_RT_PRIO 100
1569 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1571 #define MAX_PRIO (MAX_RT_PRIO + 40)
1572 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1574 static inline int rt_prio(int prio)
1576 if (unlikely(prio < MAX_RT_PRIO))
1577 return 1;
1578 return 0;
1581 static inline int rt_task(struct task_struct *p)
1583 return rt_prio(p->prio);
1586 static inline struct pid *task_pid(struct task_struct *task)
1588 return task->pids[PIDTYPE_PID].pid;
1591 static inline struct pid *task_tgid(struct task_struct *task)
1593 return task->group_leader->pids[PIDTYPE_PID].pid;
1597 * Without tasklist or rcu lock it is not safe to dereference
1598 * the result of task_pgrp/task_session even if task == current,
1599 * we can race with another thread doing sys_setsid/sys_setpgid.
1601 static inline struct pid *task_pgrp(struct task_struct *task)
1603 return task->group_leader->pids[PIDTYPE_PGID].pid;
1606 static inline struct pid *task_session(struct task_struct *task)
1608 return task->group_leader->pids[PIDTYPE_SID].pid;
1611 struct pid_namespace;
1614 * the helpers to get the task's different pids as they are seen
1615 * from various namespaces
1617 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1618 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1619 * current.
1620 * task_xid_nr_ns() : id seen from the ns specified;
1622 * set_task_vxid() : assigns a virtual id to a task;
1624 * see also pid_nr() etc in include/linux/pid.h
1626 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1627 struct pid_namespace *ns);
1629 static inline pid_t task_pid_nr(struct task_struct *tsk)
1631 return tsk->pid;
1634 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1635 struct pid_namespace *ns)
1637 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1640 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1642 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1646 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1648 return tsk->tgid;
1651 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1653 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1655 return pid_vnr(task_tgid(tsk));
1659 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1660 struct pid_namespace *ns)
1662 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1665 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1667 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1671 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1672 struct pid_namespace *ns)
1674 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1677 static inline pid_t task_session_vnr(struct task_struct *tsk)
1679 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1682 /* obsolete, do not use */
1683 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1685 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1689 * pid_alive - check that a task structure is not stale
1690 * @p: Task structure to be checked.
1692 * Test if a process is not yet dead (at most zombie state)
1693 * If pid_alive fails, then pointers within the task structure
1694 * can be stale and must not be dereferenced.
1696 static inline int pid_alive(struct task_struct *p)
1698 return p->pids[PIDTYPE_PID].pid != NULL;
1702 * is_global_init - check if a task structure is init
1703 * @tsk: Task structure to be checked.
1705 * Check if a task structure is the first user space task the kernel created.
1707 static inline int is_global_init(struct task_struct *tsk)
1709 return tsk->pid == 1;
1713 * is_container_init:
1714 * check whether in the task is init in its own pid namespace.
1716 extern int is_container_init(struct task_struct *tsk);
1718 extern struct pid *cad_pid;
1720 extern void free_task(struct task_struct *tsk);
1721 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1723 extern void __put_task_struct(struct task_struct *t);
1725 static inline void put_task_struct(struct task_struct *t)
1727 if (atomic_dec_and_test(&t->usage))
1728 __put_task_struct(t);
1731 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1732 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1735 * Per process flags
1737 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1738 /* Not implemented yet, only for 486*/
1739 #define PF_STARTING 0x00000002 /* being created */
1740 #define PF_EXITING 0x00000004 /* getting shut down */
1741 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1742 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1743 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1744 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1745 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1746 #define PF_DUMPCORE 0x00000200 /* dumped core */
1747 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1748 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1749 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1750 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1751 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1752 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1753 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1754 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1755 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1756 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1757 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1758 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1759 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1760 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1761 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1762 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1763 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1764 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1765 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1766 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1767 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1768 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1771 * Only the _current_ task can read/write to tsk->flags, but other
1772 * tasks can access tsk->flags in readonly mode for example
1773 * with tsk_used_math (like during threaded core dumping).
1774 * There is however an exception to this rule during ptrace
1775 * or during fork: the ptracer task is allowed to write to the
1776 * child->flags of its traced child (same goes for fork, the parent
1777 * can write to the child->flags), because we're guaranteed the
1778 * child is not running and in turn not changing child->flags
1779 * at the same time the parent does it.
1781 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1782 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1783 #define clear_used_math() clear_stopped_child_used_math(current)
1784 #define set_used_math() set_stopped_child_used_math(current)
1785 #define conditional_stopped_child_used_math(condition, child) \
1786 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1787 #define conditional_used_math(condition) \
1788 conditional_stopped_child_used_math(condition, current)
1789 #define copy_to_stopped_child_used_math(child) \
1790 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1791 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1792 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1793 #define used_math() tsk_used_math(current)
1795 #ifdef CONFIG_TREE_PREEMPT_RCU
1797 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1798 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1800 static inline void rcu_copy_process(struct task_struct *p)
1802 p->rcu_read_lock_nesting = 0;
1803 p->rcu_read_unlock_special = 0;
1804 p->rcu_blocked_node = NULL;
1805 INIT_LIST_HEAD(&p->rcu_node_entry);
1808 #else
1810 static inline void rcu_copy_process(struct task_struct *p)
1814 #endif
1816 #ifdef CONFIG_SMP
1817 extern int set_cpus_allowed_ptr(struct task_struct *p,
1818 const struct cpumask *new_mask);
1819 #else
1820 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1821 const struct cpumask *new_mask)
1823 if (!cpumask_test_cpu(0, new_mask))
1824 return -EINVAL;
1825 return 0;
1827 #endif
1829 #ifndef CONFIG_CPUMASK_OFFSTACK
1830 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1832 return set_cpus_allowed_ptr(p, &new_mask);
1834 #endif
1837 * Architectures can set this to 1 if they have specified
1838 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1839 * but then during bootup it turns out that sched_clock()
1840 * is reliable after all:
1842 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1843 extern int sched_clock_stable;
1844 #endif
1846 /* ftrace calls sched_clock() directly */
1847 extern unsigned long long notrace sched_clock(void);
1849 extern void sched_clock_init(void);
1850 extern u64 sched_clock_cpu(int cpu);
1852 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1853 static inline void sched_clock_tick(void)
1857 static inline void sched_clock_idle_sleep_event(void)
1861 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1864 #else
1865 extern void sched_clock_tick(void);
1866 extern void sched_clock_idle_sleep_event(void);
1867 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1868 #endif
1871 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1872 * clock constructed from sched_clock():
1874 extern unsigned long long cpu_clock(int cpu);
1876 extern unsigned long long
1877 task_sched_runtime(struct task_struct *task);
1878 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1880 /* sched_exec is called by processes performing an exec */
1881 #ifdef CONFIG_SMP
1882 extern void sched_exec(void);
1883 #else
1884 #define sched_exec() {}
1885 #endif
1887 extern void sched_clock_idle_sleep_event(void);
1888 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1890 #ifdef CONFIG_HOTPLUG_CPU
1891 extern void idle_task_exit(void);
1892 #else
1893 static inline void idle_task_exit(void) {}
1894 #endif
1896 extern void sched_idle_next(void);
1898 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1899 extern void wake_up_idle_cpu(int cpu);
1900 #else
1901 static inline void wake_up_idle_cpu(int cpu) { }
1902 #endif
1904 extern unsigned int sysctl_sched_latency;
1905 extern unsigned int sysctl_sched_min_granularity;
1906 extern unsigned int sysctl_sched_wakeup_granularity;
1907 extern unsigned int sysctl_sched_shares_ratelimit;
1908 extern unsigned int sysctl_sched_shares_thresh;
1909 extern unsigned int sysctl_sched_child_runs_first;
1911 enum sched_tunable_scaling {
1912 SCHED_TUNABLESCALING_NONE,
1913 SCHED_TUNABLESCALING_LOG,
1914 SCHED_TUNABLESCALING_LINEAR,
1915 SCHED_TUNABLESCALING_END,
1917 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1919 #ifdef CONFIG_SCHED_DEBUG
1920 extern unsigned int sysctl_sched_migration_cost;
1921 extern unsigned int sysctl_sched_nr_migrate;
1922 extern unsigned int sysctl_sched_time_avg;
1923 extern unsigned int sysctl_timer_migration;
1925 int sched_proc_update_handler(struct ctl_table *table, int write,
1926 void __user *buffer, size_t *length,
1927 loff_t *ppos);
1928 #endif
1929 #ifdef CONFIG_SCHED_DEBUG
1930 static inline unsigned int get_sysctl_timer_migration(void)
1932 return sysctl_timer_migration;
1934 #else
1935 static inline unsigned int get_sysctl_timer_migration(void)
1937 return 1;
1939 #endif
1940 extern unsigned int sysctl_sched_rt_period;
1941 extern int sysctl_sched_rt_runtime;
1943 int sched_rt_handler(struct ctl_table *table, int write,
1944 void __user *buffer, size_t *lenp,
1945 loff_t *ppos);
1947 extern unsigned int sysctl_sched_compat_yield;
1949 #ifdef CONFIG_RT_MUTEXES
1950 extern int rt_mutex_getprio(struct task_struct *p);
1951 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1952 extern void rt_mutex_adjust_pi(struct task_struct *p);
1953 #else
1954 static inline int rt_mutex_getprio(struct task_struct *p)
1956 return p->normal_prio;
1958 # define rt_mutex_adjust_pi(p) do { } while (0)
1959 #endif
1961 extern void set_user_nice(struct task_struct *p, long nice);
1962 extern int task_prio(const struct task_struct *p);
1963 extern int task_nice(const struct task_struct *p);
1964 extern int can_nice(const struct task_struct *p, const int nice);
1965 extern int task_curr(const struct task_struct *p);
1966 extern int idle_cpu(int cpu);
1967 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1968 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1969 struct sched_param *);
1970 extern struct task_struct *idle_task(int cpu);
1971 extern struct task_struct *curr_task(int cpu);
1972 extern void set_curr_task(int cpu, struct task_struct *p);
1974 void yield(void);
1977 * The default (Linux) execution domain.
1979 extern struct exec_domain default_exec_domain;
1981 union thread_union {
1982 struct thread_info thread_info;
1983 unsigned long stack[THREAD_SIZE/sizeof(long)];
1986 #ifndef __HAVE_ARCH_KSTACK_END
1987 static inline int kstack_end(void *addr)
1989 /* Reliable end of stack detection:
1990 * Some APM bios versions misalign the stack
1992 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1994 #endif
1996 extern union thread_union init_thread_union;
1997 extern struct task_struct init_task;
1999 extern struct mm_struct init_mm;
2001 extern struct pid_namespace init_pid_ns;
2004 * find a task by one of its numerical ids
2006 * find_task_by_pid_ns():
2007 * finds a task by its pid in the specified namespace
2008 * find_task_by_vpid():
2009 * finds a task by its virtual pid
2011 * see also find_vpid() etc in include/linux/pid.h
2014 extern struct task_struct *find_task_by_vpid(pid_t nr);
2015 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2016 struct pid_namespace *ns);
2018 extern void __set_special_pids(struct pid *pid);
2020 /* per-UID process charging. */
2021 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2022 static inline struct user_struct *get_uid(struct user_struct *u)
2024 atomic_inc(&u->__count);
2025 return u;
2027 extern void free_uid(struct user_struct *);
2028 extern void release_uids(struct user_namespace *ns);
2030 #include <asm/current.h>
2032 extern void do_timer(unsigned long ticks);
2034 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2035 extern int wake_up_process(struct task_struct *tsk);
2036 extern void wake_up_new_task(struct task_struct *tsk,
2037 unsigned long clone_flags);
2038 #ifdef CONFIG_SMP
2039 extern void kick_process(struct task_struct *tsk);
2040 #else
2041 static inline void kick_process(struct task_struct *tsk) { }
2042 #endif
2043 extern void sched_fork(struct task_struct *p, int clone_flags);
2044 extern void sched_dead(struct task_struct *p);
2046 extern void proc_caches_init(void);
2047 extern void flush_signals(struct task_struct *);
2048 extern void __flush_signals(struct task_struct *);
2049 extern void ignore_signals(struct task_struct *);
2050 extern void flush_signal_handlers(struct task_struct *, int force_default);
2051 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2053 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2055 unsigned long flags;
2056 int ret;
2058 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2059 ret = dequeue_signal(tsk, mask, info);
2060 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2062 return ret;
2065 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2066 sigset_t *mask);
2067 extern void unblock_all_signals(void);
2068 extern void release_task(struct task_struct * p);
2069 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2070 extern int force_sigsegv(int, struct task_struct *);
2071 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2072 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2073 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2074 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2075 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2076 extern int kill_pid(struct pid *pid, int sig, int priv);
2077 extern int kill_proc_info(int, struct siginfo *, pid_t);
2078 extern int do_notify_parent(struct task_struct *, int);
2079 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2080 extern void force_sig(int, struct task_struct *);
2081 extern void force_sig_specific(int, struct task_struct *);
2082 extern int send_sig(int, struct task_struct *, int);
2083 extern void zap_other_threads(struct task_struct *p);
2084 extern struct sigqueue *sigqueue_alloc(void);
2085 extern void sigqueue_free(struct sigqueue *);
2086 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2087 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2088 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2090 static inline int kill_cad_pid(int sig, int priv)
2092 return kill_pid(cad_pid, sig, priv);
2095 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2096 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2097 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2098 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2100 static inline int is_si_special(const struct siginfo *info)
2102 return info <= SEND_SIG_FORCED;
2106 * True if we are on the alternate signal stack.
2108 static inline int on_sig_stack(unsigned long sp)
2110 #ifdef CONFIG_STACK_GROWSUP
2111 return sp >= current->sas_ss_sp &&
2112 sp - current->sas_ss_sp < current->sas_ss_size;
2113 #else
2114 return sp > current->sas_ss_sp &&
2115 sp - current->sas_ss_sp <= current->sas_ss_size;
2116 #endif
2119 static inline int sas_ss_flags(unsigned long sp)
2121 return (current->sas_ss_size == 0 ? SS_DISABLE
2122 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2126 * Routines for handling mm_structs
2128 extern struct mm_struct * mm_alloc(void);
2130 /* mmdrop drops the mm and the page tables */
2131 extern void __mmdrop(struct mm_struct *);
2132 static inline void mmdrop(struct mm_struct * mm)
2134 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2135 __mmdrop(mm);
2138 /* mmput gets rid of the mappings and all user-space */
2139 extern void mmput(struct mm_struct *);
2140 /* Grab a reference to a task's mm, if it is not already going away */
2141 extern struct mm_struct *get_task_mm(struct task_struct *task);
2142 /* Remove the current tasks stale references to the old mm_struct */
2143 extern void mm_release(struct task_struct *, struct mm_struct *);
2144 /* Allocate a new mm structure and copy contents from tsk->mm */
2145 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2147 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2148 struct task_struct *, struct pt_regs *);
2149 extern void flush_thread(void);
2150 extern void exit_thread(void);
2152 extern void exit_files(struct task_struct *);
2153 extern void __cleanup_signal(struct signal_struct *);
2154 extern void __cleanup_sighand(struct sighand_struct *);
2156 extern void exit_itimers(struct signal_struct *);
2157 extern void flush_itimer_signals(void);
2159 extern NORET_TYPE void do_group_exit(int);
2161 extern void daemonize(const char *, ...);
2162 extern int allow_signal(int);
2163 extern int disallow_signal(int);
2165 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2166 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2167 struct task_struct *fork_idle(int);
2169 extern void set_task_comm(struct task_struct *tsk, char *from);
2170 extern char *get_task_comm(char *to, struct task_struct *tsk);
2172 #ifdef CONFIG_SMP
2173 extern void wait_task_context_switch(struct task_struct *p);
2174 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2175 #else
2176 static inline void wait_task_context_switch(struct task_struct *p) {}
2177 static inline unsigned long wait_task_inactive(struct task_struct *p,
2178 long match_state)
2180 return 1;
2182 #endif
2184 #define next_task(p) \
2185 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2187 #define for_each_process(p) \
2188 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2190 extern bool current_is_single_threaded(void);
2193 * Careful: do_each_thread/while_each_thread is a double loop so
2194 * 'break' will not work as expected - use goto instead.
2196 #define do_each_thread(g, t) \
2197 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2199 #define while_each_thread(g, t) \
2200 while ((t = next_thread(t)) != g)
2202 /* de_thread depends on thread_group_leader not being a pid based check */
2203 #define thread_group_leader(p) (p == p->group_leader)
2205 /* Do to the insanities of de_thread it is possible for a process
2206 * to have the pid of the thread group leader without actually being
2207 * the thread group leader. For iteration through the pids in proc
2208 * all we care about is that we have a task with the appropriate
2209 * pid, we don't actually care if we have the right task.
2211 static inline int has_group_leader_pid(struct task_struct *p)
2213 return p->pid == p->tgid;
2216 static inline
2217 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2219 return p1->tgid == p2->tgid;
2222 static inline struct task_struct *next_thread(const struct task_struct *p)
2224 return list_entry_rcu(p->thread_group.next,
2225 struct task_struct, thread_group);
2228 static inline int thread_group_empty(struct task_struct *p)
2230 return list_empty(&p->thread_group);
2233 #define delay_group_leader(p) \
2234 (thread_group_leader(p) && !thread_group_empty(p))
2236 static inline int task_detached(struct task_struct *p)
2238 return p->exit_signal == -1;
2242 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2243 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2244 * pins the final release of task.io_context. Also protects ->cpuset and
2245 * ->cgroup.subsys[].
2247 * Nests both inside and outside of read_lock(&tasklist_lock).
2248 * It must not be nested with write_lock_irq(&tasklist_lock),
2249 * neither inside nor outside.
2251 static inline void task_lock(struct task_struct *p)
2253 spin_lock(&p->alloc_lock);
2256 static inline void task_unlock(struct task_struct *p)
2258 spin_unlock(&p->alloc_lock);
2261 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2262 unsigned long *flags);
2264 static inline void unlock_task_sighand(struct task_struct *tsk,
2265 unsigned long *flags)
2267 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2270 #ifndef __HAVE_THREAD_FUNCTIONS
2272 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2273 #define task_stack_page(task) ((task)->stack)
2275 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2277 *task_thread_info(p) = *task_thread_info(org);
2278 task_thread_info(p)->task = p;
2281 static inline unsigned long *end_of_stack(struct task_struct *p)
2283 return (unsigned long *)(task_thread_info(p) + 1);
2286 #endif
2288 static inline int object_is_on_stack(void *obj)
2290 void *stack = task_stack_page(current);
2292 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2295 extern void thread_info_cache_init(void);
2297 #ifdef CONFIG_DEBUG_STACK_USAGE
2298 static inline unsigned long stack_not_used(struct task_struct *p)
2300 unsigned long *n = end_of_stack(p);
2302 do { /* Skip over canary */
2303 n++;
2304 } while (!*n);
2306 return (unsigned long)n - (unsigned long)end_of_stack(p);
2308 #endif
2310 /* set thread flags in other task's structures
2311 * - see asm/thread_info.h for TIF_xxxx flags available
2313 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2315 set_ti_thread_flag(task_thread_info(tsk), flag);
2318 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2320 clear_ti_thread_flag(task_thread_info(tsk), flag);
2323 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2325 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2328 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2330 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2333 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2335 return test_ti_thread_flag(task_thread_info(tsk), flag);
2338 static inline void set_tsk_need_resched(struct task_struct *tsk)
2340 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2343 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2345 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2348 static inline int test_tsk_need_resched(struct task_struct *tsk)
2350 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2353 static inline int restart_syscall(void)
2355 set_tsk_thread_flag(current, TIF_SIGPENDING);
2356 return -ERESTARTNOINTR;
2359 static inline int signal_pending(struct task_struct *p)
2361 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2364 static inline int __fatal_signal_pending(struct task_struct *p)
2366 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2369 static inline int fatal_signal_pending(struct task_struct *p)
2371 return signal_pending(p) && __fatal_signal_pending(p);
2374 static inline int signal_pending_state(long state, struct task_struct *p)
2376 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2377 return 0;
2378 if (!signal_pending(p))
2379 return 0;
2381 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2384 static inline int need_resched(void)
2386 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2390 * cond_resched() and cond_resched_lock(): latency reduction via
2391 * explicit rescheduling in places that are safe. The return
2392 * value indicates whether a reschedule was done in fact.
2393 * cond_resched_lock() will drop the spinlock before scheduling,
2394 * cond_resched_softirq() will enable bhs before scheduling.
2396 extern int _cond_resched(void);
2398 #define cond_resched() ({ \
2399 __might_sleep(__FILE__, __LINE__, 0); \
2400 _cond_resched(); \
2403 extern int __cond_resched_lock(spinlock_t *lock);
2405 #ifdef CONFIG_PREEMPT
2406 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2407 #else
2408 #define PREEMPT_LOCK_OFFSET 0
2409 #endif
2411 #define cond_resched_lock(lock) ({ \
2412 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2413 __cond_resched_lock(lock); \
2416 extern int __cond_resched_softirq(void);
2418 #define cond_resched_softirq() ({ \
2419 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2420 __cond_resched_softirq(); \
2424 * Does a critical section need to be broken due to another
2425 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2426 * but a general need for low latency)
2428 static inline int spin_needbreak(spinlock_t *lock)
2430 #ifdef CONFIG_PREEMPT
2431 return spin_is_contended(lock);
2432 #else
2433 return 0;
2434 #endif
2438 * Thread group CPU time accounting.
2440 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2441 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2443 static inline void thread_group_cputime_init(struct signal_struct *sig)
2445 sig->cputimer.cputime = INIT_CPUTIME;
2446 spin_lock_init(&sig->cputimer.lock);
2447 sig->cputimer.running = 0;
2450 static inline void thread_group_cputime_free(struct signal_struct *sig)
2455 * Reevaluate whether the task has signals pending delivery.
2456 * Wake the task if so.
2457 * This is required every time the blocked sigset_t changes.
2458 * callers must hold sighand->siglock.
2460 extern void recalc_sigpending_and_wake(struct task_struct *t);
2461 extern void recalc_sigpending(void);
2463 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2466 * Wrappers for p->thread_info->cpu access. No-op on UP.
2468 #ifdef CONFIG_SMP
2470 static inline unsigned int task_cpu(const struct task_struct *p)
2472 return task_thread_info(p)->cpu;
2475 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2477 #else
2479 static inline unsigned int task_cpu(const struct task_struct *p)
2481 return 0;
2484 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2488 #endif /* CONFIG_SMP */
2490 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2492 #ifdef CONFIG_TRACING
2493 extern void
2494 __trace_special(void *__tr, void *__data,
2495 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2496 #else
2497 static inline void
2498 __trace_special(void *__tr, void *__data,
2499 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2502 #endif
2504 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2505 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2507 extern void normalize_rt_tasks(void);
2509 #ifdef CONFIG_GROUP_SCHED
2511 extern struct task_group init_task_group;
2512 #ifdef CONFIG_USER_SCHED
2513 extern struct task_group root_task_group;
2514 extern void set_tg_uid(struct user_struct *user);
2515 #endif
2517 extern struct task_group *sched_create_group(struct task_group *parent);
2518 extern void sched_destroy_group(struct task_group *tg);
2519 extern void sched_move_task(struct task_struct *tsk);
2520 #ifdef CONFIG_FAIR_GROUP_SCHED
2521 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2522 extern unsigned long sched_group_shares(struct task_group *tg);
2523 #endif
2524 #ifdef CONFIG_RT_GROUP_SCHED
2525 extern int sched_group_set_rt_runtime(struct task_group *tg,
2526 long rt_runtime_us);
2527 extern long sched_group_rt_runtime(struct task_group *tg);
2528 extern int sched_group_set_rt_period(struct task_group *tg,
2529 long rt_period_us);
2530 extern long sched_group_rt_period(struct task_group *tg);
2531 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2532 #endif
2533 #endif
2535 extern int task_can_switch_user(struct user_struct *up,
2536 struct task_struct *tsk);
2538 #ifdef CONFIG_TASK_XACCT
2539 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2541 tsk->ioac.rchar += amt;
2544 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2546 tsk->ioac.wchar += amt;
2549 static inline void inc_syscr(struct task_struct *tsk)
2551 tsk->ioac.syscr++;
2554 static inline void inc_syscw(struct task_struct *tsk)
2556 tsk->ioac.syscw++;
2558 #else
2559 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2563 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2567 static inline void inc_syscr(struct task_struct *tsk)
2571 static inline void inc_syscw(struct task_struct *tsk)
2574 #endif
2576 #ifndef TASK_SIZE_OF
2577 #define TASK_SIZE_OF(tsk) TASK_SIZE
2578 #endif
2581 * Call the function if the target task is executing on a CPU right now:
2583 extern void task_oncpu_function_call(struct task_struct *p,
2584 void (*func) (void *info), void *info);
2587 #ifdef CONFIG_MM_OWNER
2588 extern void mm_update_next_owner(struct mm_struct *mm);
2589 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2590 #else
2591 static inline void mm_update_next_owner(struct mm_struct *mm)
2595 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2598 #endif /* CONFIG_MM_OWNER */
2600 #endif /* __KERNEL__ */
2602 #endif